Method and system to monitor pipeline condition

Abstract

A method and system for real time monitoring of the condition of a pipeline is disclosed. The method comprises continuously monitoring transient pressure information of a fluid in the pipeline, selecting a time window of transient pressure information and processing the time window of transient pressure information to detect an anomaly in the pipeline.

The invention claimed is: 
     
       1. A method for real time monitoring of a condition of a pipeline, comprising:
 continuously monitoring transient pressure information of a fluid in the pipeline; 
 selecting a time window of transient pressure information; and 
 processing the time window of transient pressure information to detect an anomaly in the pipeline, wherein processing the time window of transient pressure information comprises: 
 down sampling the time window of transient pressure information to generate a down sampled time window of pressure information; and 
 processing the down sampled time window of transient pressure information by a classifier artificial neural network (ANN) trained to categorise a pipeline operating condition, wherein a size of the down sampled time window of transient pressure information has a same size as an input to the classifier ANN; and 
 determining by the classifier ANN the pipeline operating condition for the time window. 
 
     
     
       2. The method of  claim 1 , wherein the pipeline operating condition is categorised as a normal operating condition where no anomaly is present in the time window of transient pressure information or an abnormal pressure condition where an anomaly is present in the time window of transient pressure information. 
     
     
       3. The method of  claim 2 , wherein the category of abnormal pressure condition is further categorised into a first abnormal pressure condition signifying only a presence of an anomaly in the time window of transient pressure information or a second abnormal condition where not only the presence of an anomaly is detected in the time window of transient pressure information but also that anomaly characteristics related to the anomaly in the time window of transient pressure information may be determined. 
     
     
       4. The method of  claim 3 , wherein processing the time window of transient pressure information comprises:
 following determining that the time window of transient pressure information is categorised in the second abnormal condition; then 
 processing the down sampled time window of transient pressure information by a first anomaly detector ANN trained to detect a first type of anomaly in the pipeline and to determine associated anomaly characteristics for the first type of anomaly; and 
 verifying whether the first type of anomaly is detected in the time window of transient pressure information. 
 
     
     
       5. The method of  claim 4 , wherein verifying whether the first type of anomaly is detected in the time window of transient pressure information comprises determining whether the determined associated anomaly characteristics are consistent with the pipeline. 
     
     
       6. The method of  claim 5 , wherein determining whether the determined associated anomaly characteristics are consistent with the pipeline includes determining whether a location of the anomaly is consistent with a length of pipeline. 
     
     
       7. The method of  claim 5 , wherein on determining that the determined associated anomaly characteristics are consistent with the pipeline, verifying whether the first type of anomaly is detected in the time window of transient pressure information further comprises:
 numerically generating a time window of theoretical pressure information based on the first anomaly type and associated anomaly characteristics; 
 comparing the selected time window of transient pressure information as measured with the time window of theoretical pressure information as numerically generated to determine a comparison measure; and 
 applying a comparison threshold to the comparison measure to indicate that the first type of anomaly is detected in the time window of transient pressure information. 
 
     
     
       8. The method of  claim 7 , wherein on determining that the numerically generated time window of transient pressure information is partially consistent with the selected time window of transient pressure information the method further comprises:
 refining one or more anomaly characteristics to obtain a better match between the numerically generated time window of transient pressure information and the selected time window of transient pressure information. 
 
     
     
       9. The method of  claim 4 , wherein on failing to verify that the first type of anomaly is detected in the time window of transient pressure information the method comprises processing the down sampled time window of transient pressure information by a second anomaly detector ANN trained to detect a second type of anomaly in the pipeline and to determine associated anomaly characteristics for the second type of anomaly. 
     
     
       10. The method of  claim 3 , wherein processing the time window of transient pressure information comprises following determining that the time window of transient pressure information is categorised in the second abnormal condition then successively selecting a number of time windows of transient pressure information following in time from the selected time window of transient pressure information covering a predetermined time period and determining that all time windows are categorised in the second abnormal condition. 
     
     
       11. The method of  claim 1 , wherein following processing of the time window of transient pressure information a further successive time window of transient pressure information is selected for processing. 
     
     
       12. The method of  claim 1 , wherein the anomaly include:
 a leak in the pipeline; 
 a burst in the pipeline; 
 a closure or opening of a valve in the pipeline; or 
 a non-characteristic flow event. 
 
     
     
       13. The method of  claim 1 , wherein characteristics associated with the anomaly include:
 a location of the anomaly; 
 a physical size of the anomaly; or 
 a flow rate of liquid associated with the anomaly. 
 
     
     
       14. The method of  claim 1 , wherein the classifier ANN is trained to categorise a pipeline operating condition by:
 selecting a range of anomaly types and associated ranges of anomaly characteristics; 
 generating respective time windows of transient pressure information for the range of anomaly types and associated ranges of values of the anomaly characteristics; 
 down sampling the respective time windows of transient pressure information to form respective down sampled time windows of transient pressure information each having a size corresponding to the size of the input of the classifier ANN; 
 assigning a respective pipeline operating condition to each of the respective down sampled time windows of transient pressure information; and 
 training the classifier ANN to determine the pipeline operating condition based on each of the respective down sampled time windows of pressure information and the assigned respective pipeline operating condition. 
 
     
     
       15. The method of  claim 14 , wherein the classifier ANN is trained to determine a respective anomaly type and associated anomaly characteristics by:
 training the classifier ANN to determine the respective anomaly type and associated anomaly characteristics based on each of the respective down sampled time windows of transient pressure information and their anomaly type and anomaly characteristics. 
 
     
     
       16. The method of  claim 14 , wherein generating respective time windows of transient pressure information for the range of anomaly types and associated ranges of values of the anomaly characteristics includes numerically generating one or more of the respective time windows of transient pressure information based on a hydrodynamic model of the pipeline. 
     
     
       17. The method of  claim 14 , wherein generating respective time windows of transient pressure information for the range of anomaly types and associated ranges of values of the anomaly characteristics includes basing one or more of the respective time windows of transient pressure information on historical pressure information. 
     
     
       18. A system for real time monitoring of the condition of a pipeline, the system including:
 a pressure detector for monitoring transient pressure information of a fluid in the pipeline; 
 an analysis module comprising one or more data processors for: 
 selecting a time window of transient pressure information; and 
 processing the time window of transient pressure information to detect an anomaly in the pipeline, wherein processing the time window of transient pressure information by the analysis module comprises: 
 down sampling the time window of transient pressure information to generate a down sampled time window of pressure information; and 
 processing the down sampled time window of transient pressure information by a classifier artificial neural network (ANN) trained to categorise a pipeline operating condition, wherein a size of the down sampled time window of transient pressure information has a same size as an input to the classifier ANN; and 
 determining by the classifier ANN the pipeline operating condition for the time window. 
 
     
     
       19. The system of  claim 18 , wherein the pipeline operating condition is categorised as a normal operating condition where no anomaly is present in the time window of transient pressure information or an abnormal pressure condition where an anomaly is present in the time window of transient pressure information. 
     
     
       20. The system of  claim 19 , wherein the category of abnormal pressure condition is further categorised into a first abnormal pressure condition signifying only a presence of an anomaly in the time window of transient pressure information or a second abnormal condition where not only the presence of an anomaly is detected in the time window of transient pressure information but also that anomaly characteristics related to the anomaly in the time window of transient pressure information may be determined. 
     
     
       21. The system of  claim 20 , wherein processing the time window of transient pressure information by the analysis module comprises:
 following determining that the time window of transient pressure information is categorised in the second abnormal condition; then 
 processing the down sampled time window of transient pressure information by a first anomaly detector ANN trained to detect a first type of anomaly in the pipeline and to determine associated anomaly characteristics for the first type of anomaly; and 
 verifying whether the first type of anomaly is detected in the time window of transient pressure information. 
 
     
     
       22. The system of  claim 21 , wherein verifying whether the first type of anomaly is detected in the time window of transient pressure information comprises determining whether the determined associated anomaly characteristics are consistent with the pipeline. 
     
     
       23. The system of  claim 22 , wherein determining whether the determined associated anomaly characteristics are consistent with the pipeline includes determining whether a location of the anomaly is consistent with a length of pipeline. 
     
     
       24. The system of  claim 22 , wherein on determining that the determined associated anomaly characteristics are consistent with the pipeline, verifying whether the first type of anomaly is detected in the time window of transient pressure information further comprises:
 numerically generating a time window of theoretical pressure information based on the first anomaly type and associated anomaly characteristics; 
 comparing the selected time window of transient pressure information as measured with the time window of theoretical pressure information as numerically generated to determine a comparison measure; and 
 applying a comparison threshold to the comparison measure to indicate that the first type of anomaly is detected in the time window of transient pressure information. 
 
     
     
       25. The system of  claim 24 , wherein on determining that the numerically generated time window of transient pressure information is partially consistent with the selected time window of transient pressure information the system further comprises:
 refining one or more anomaly characteristics to obtain a better match between the numerically generated time window of transient pressure information and the selected time window of transient pressure information. 
 
     
     
       26. The system of  claim 21 , wherein on failing to verify that the first type of anomaly is detected in the time window of transient pressure information the system comprises processing the down sampled time window of transient pressure information by a second anomaly detector ANN trained to detect a second type of anomaly in the pipeline and to determine associated anomaly characteristics for the second type of anomaly. 
     
     
       27. The system of  claim 20 , wherein processing the time window of transient pressure information comprises following determining that the time window of transient pressure information is categorised in the second abnormal condition then successively selecting a number of time windows of transient pressure information following in time from the selected time window of transient pressure information covering a predetermined time period and determining that all time windows are categorised in the second abnormal condition. 
     
     
       28. The system of  claim 18 , wherein following processing of the time window of transient pressure information a further successive time window of transient pressure information is selected for processing by the analysis module. 
     
     
       29. The system of  claim 18 , wherein the anomaly include:
 a leak in the pipeline; 
 a burst in the pipeline; 
 a closure or opening of a valve in the pipeline; or 
 a non-characteristic flow event. 
 
     
     
       30. The system of  claim 18 , wherein characteristics associated with the anomaly include:
 a location of the anomaly; 
 a physical size of the anomaly; or 
 a flow rate of liquid associated with the anomaly. 
 
     
     
       31. The system of  claim 18 , wherein the classifier ANN is trained to categorise a pipeline operating condition by:
 selecting a range of anomaly types and associated ranges of anomaly characteristics; 
 generating respective time windows of transient pressure information for the range of anomaly types and associated ranges of values of the anomaly characteristics; 
 down sampling the respective time windows of transient pressure information to form respective down sampled time windows of transient pressure information each having a size corresponding to the size of the input of the classifier ANN; 
 assigning a respective pipeline operating condition to each of the respective down sampled time windows of transient pressure information; and 
 training the classifier ANN to determine the pipeline operating condition based on each of the respective down sampled time windows of pressure information and the assigned respective pipeline operating condition. 
 
     
     
       32. The system of  claim 31 , wherein the classifier ANN is trained to determine a respective anomaly type and associated anomaly characteristics by:
 training the classifier ANN to determine the respective anomaly type and associated anomaly characteristics based on each of the respective down sampled time windows of transient pressure information and their anomaly type and anomaly characteristics. 
 
     
     
       33. The system of  claim 31 , wherein generating respective time windows of transient pressure information for the range of anomaly types and associated ranges of values of the anomaly characteristics includes numerically generating one or more of the respective time windows of transient pressure information based on a hydrodynamic model of the pipeline. 
     
     
       34. The system of  claim 31 , wherein generating respective time windows of transient pressure information for the range of anomaly types and associated ranges of values of the anomaly characteristics includes basing one or more of the respective time windows of transient pressure information on historical pressure information.